top of page
  • jamieledingham

The curious case of the high BFIHOST

Having a read through the new CIRIA NFM manual recently, I was reminded of some interesting hydrological characteristics exhibited on a small site I've been looking at over in the east of Scotland. Prompted in particular by the emphasis the manual places on 'understanding your catchment', this post looks at integrating multiple sources of apparently initially conflicting information to try and characterise small catchment hydrology.

Due to the relatively early stages of the project, the exact location will remain nameless for now. It's in north-east Scotland and is relatively typical of the local area in terms of land-use. Interest was first prompted by the BFIHOST value of 0.83; working mostly in Scotland it's relatively rare to see such high numbers. While the country isn't bereft of catchments with reasonable baseflow contributions, locally important groundwater sources or productive Aquifers I've yet to find a higher index in this region; in the area in question local rivers such as the Ugie, Ythan, and Don all have BF indices in the 0.58-0.65 region.

A number of initial questions came to mind once I first identified the value, but for this post I'm going to concentrate on:

  1. Is the high BFIHOST value 'right' and how does it reconcile with other observations and evidence?

  2. What have I learnt from the process of integrating multiple evidence sources to assess this?

Is it 'right' and does it reconcile with other observations and evidence?

By 'right' I mean whether the proportion of sub-surface flow relative to quick runoff is correctly indexed by BFIHOST.

In favour of supporting a high BFI are the gauged local rivers; while they don't exhibit such high calculated BF values as the study site, they tend to show a greater bias towards slower, sub-surface contributions over faster surface contributions (BF values >0.5). At the larger catchment scale it's reasonable to expect that both high and low BFI values get moderated; it's less likely in this part of the country that larger catchments are entirely dominated by a single drift lithology so extremes tend to get averaged out. The site in question is several magnitude smaller and so has greater potential to be influenced by small-scale changes in drift geology.

Historically the local area has made significant use of wells as a water supply; this supports the idea of the drift geology having some productive water-bearing capability.

Against this are local observations of the site over a number of years suggest extremely high runoff coefficients during certain periods; there are numerous examples of overland flow and ponding during storms.

Squelch, Squelch. The abundance of rush confirming what my leaking welly has already told me: that the site is very wet (even after a few days of no rain).

Infiltration as a drainage management option also has a very mixed local success rate.

At first glance the available information appeared, to some extent, contradictory. On further review, thought and discussion with a couple of colleagues, my thoughts now lie along the following lines.

It's possible (probable?) that the drift geology and soils give rise to hydrological behaviour that shows significant temporal and spatial variation. While this is true of any UK catchment, in this case it helps to explains why there is such an apparent discrepancy between the observed high levels of runoff and the high BFIHOST value. BFI/BFIHOST characterises runoff as a single value; in practice the proportion of runoff from faster and slower pathways is probably dependent on the prevailing hydrological conditions such as antecedent rainfall and soil moisture. In conjunction with this, I'm aware that certain areas of the catchment exhibit variable infiltration characteristics.

The current working theory is that that the clay like topsoil limits (but does not entirely prevent) infiltration and is variable in both its extent and thickness. Underlying this clay like topsoil are areas of sands and gravels of moderate productivity. This explanation is currently the most plausible of the (relatively few!) working theories that reconcile with the available evidence.

Integrating multiple evidence sources.

One of the reasons hydrology interests me is because it's not straightforward; particularly when dealing with the sub-surface there are still severe limitations in understanding how water moves through the drift geology. Similar to how avalanche pits are often only useful for telling you the risk for the 1m square pit site you've excavated, site investigation only ever gives a brief window into what's happening with water movement as there are often rapid changes in characteristics over both space and time. Developing even a basic conceptual understanding of water movement requires analysts to use diverse sources of information, some of them potentially 'unscientific'. Available information for this work included:

  • Site observations and photographs from residents;

  • Soakaway test results;

  • Photographs of foundation excavations;

  • BGS borehole records

  • Botanical survey interpretation;

  • Soil and geology drift mapping;

  • Aerial photography

  • Electrical conductivity measurements

  • Walkover observations including topsoil assessment.

The project under consideration here is not unusual and runoff management at source is generally a topic of interest at present. There are a range of investigations available to the hydrologist; at one end of the scale a cursory look at the map and acceptance of BFIHOST type values and at the other an academic type approach involving significant catchment instrumentation over a number of years and possibly the development of distributed, physically-based rainfall-runoff modelling. While it's fairly evident the resources to undertake the latter type approach are rarely available for small applied projects, neither is it justifiable to only rely on broad-scale mapping and indices. It is for this reason that this post uses an example of the 'middle ground' as it's where I perceive the majority of applied work is undertaken.

What lies beneath... It's rare to get good insights into what's going on below ground, even cuttings like these only give a snapshot in time and space.

The broader observations I'd draw from this work are:

  1. Local observations of sites can be invaluable as they can often be built up from long periods of observations and the pattern interpretation has already been done for you (to some extent). Without local observation on this site it would have been difficult to identify the variable nature of the runoff characteristics in the first place, never mind explain them. Clearly there is still a risk of observational/observer bias but in with that in mind the advantages for the analyst can often outweigh the disadvantages. With the move towards citizen science and broader stakeholder involvement in flood risk management decisions, scientific approaches to hydrology need to become comfortable with integrating more qualitative (but often incredibly insightful) observations.

  2. It's easy to characterise catchments using numbers like BFIHOST and PROPWET and these often do give a good general idea of behaviour. However, for certain projects in certain locations, there needs to be considerably more scrutiny using these indexes; this is particularly the case where runoff is proposed to be managed on a small scale at or near it's source and therefore understanding runoff generation becomes critical to the success of any intervention.

  3. Indexes often give a general picture; for example BFI or HOSTBFI indexes across a hydrograph. Event or seasonal characteristics may be significantly different, so if interventions are targeting events or problems that manifest themselves seasonally, then index suitability needs assessed (this problem equally applies when observations are limited to specific times of year or specific events).

  4. Confirming indexes using soil mapping is generally inappropriate unless highly detailed mapping from site investigation is available. HOST mapping uses soil mapping as it's base information so you end up essentially comparing the base data to the processed base data, rather than the processed base data against reality.

Work continues to refine our understanding of the site in question.


Recent Posts

See All

The SEPA time-series data service

Firstly, hats off to the person/people at SEPA who've pushed this through. For anyone who isn't aware, this is a way of retrieving time-series data held by SEPA. Of note is that is gives access to hig


bottom of page